Using Two-Photon Excitation to Control Bending ... - Semantic Scholar
full papers Photomechanical control
Using Two-Photon Excitation to Control Bending Motions in Molecular-Crystal Nanorods Jacob T. Good, Jonathan J. Burdett, and Christopher J. Bardeen*
Molecular-crystal nanorods composed of 9-anthracenecarboxylic acid can undergo reversible bending due to molecular-level geometry changes associated with the photodimerization of the molecules in the crystal lattice. The use of highly focused near-IR femtosecond laser pulses results in twophoton excitation of micrometer-scale regions and is used to induce transient bends at various locations along the length of a single 200-nm-diameter nanorod. Bending can be observed in nanorods with diameters as small as 35 nm, and translational motion of a single nanorod could be induced by sequential bending of longer segments. A kinetic model is presented that quantitatively describes the bending and relaxation dynamics of individual rods. The results of this work show that it is possible to use laser excitation conditions to control the location, rate, and magnitude of photodeformations in these nanorods. The ability to control the motion of these ultrasmall photomechanical structures may be useful for manipulating objects on the nanoscale.
1. Introduction The manipulation of nano- and micrometer scale objects is a central goal in many areas of science, from physics to biology. Photons are ideal for controlling motion since they can penetrate deep into 3D samples like biological media and do not require physical contact between the controller and object. To convert the photons into mechanical work, small-scale photomechanical actuators are required. The development of photomechanical actuators has centered on the use of photochemically active chromophores, like azobenzene and spiropyran, embedded in polymers and liquid-crystal elastomers.[1,2] Progress in this field has been rapid, with recent demonstrations of artificial inchworms, swimmers, motors, and oscillators.[3–8] Recently, a new type of organic photomechanical material, based on single-component photochemically active molecular crystals, has been demonstrated. Diarylethene
[] Prof. C. J. Bardeen, J. T. Good, J. J. Burdett Department of Chemistry University of California at Riverside Riverside, CA 92521 (USA) E-mail: [email protected] DOI: 10.1002/smll.200900895
2902
Keywords:
bending motions nanorods photomechanical control two-photon excitation
microcrystals and thin crystal plates have been shown to undergo large deformations driven by intramolecular ringopening and closing reactions activated by photons of different wavelengths,[9–12] while Cu-TCNQ (TCNQ ¼ 7,7,8,8-tetracyanoquinodimethane) crystals have also shown reversible deformations due to photoinduced charge transfer.[13,14] Our group has investigated the behavior of molecular crystalline nanorods composed of anthracene derivatives that undergo a 4þ4 photodimerization reaction. In this type of reaction, a single anthracene absorbs a photon and reacts with a neighboring molecule to form a bridged dimer, as illustrated in Figure 1.[15–17] The resulting photodimer is typically stable unless dissociation is induced by a UV photon or heat. The nanoscale dimensions of rods composed of these anthracene derivatives permit them to undergo crystal-to-crystal photoreactions that would cause a macroscopic crystal to disintegrate.[18,19] In the case of 9-anthracene carboxylic acid (9AC), the photodimer spontaneously dissociates within a few minutes, resetting the system.[20] The initial demonstrations of reversible photomechanical molecular crystals provide evidence that they could be useful as small-scale photoactuators. However, to assess whether the nanorod photoresponse can be harnessed to generate controlled, microscale displacements, several aspects of their photomechanical response must be understood in more detail.
ß 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2009, 5, No. 24, 2902–2909
Bending Motions in Molecular Crystal Nanorods
H
H
H
O
O
H H
H O
O
O
O
H O
O
O
O H
O
O
H
H
H
O H
O
O O
O
H O
O
O
O
O
O
O
H
H
H O
O H O
O
O
H
H
H
H
H
OH O
O
O
O O
O
O
H O
O
O
O
H
H
H
O
O
O
H O
O
O
O H
O
O
O
O
H
O
O
O
O
O
O
H
H O
O O
O
O
O
O H
O H
O
H
O
H
O
O
O
O
kT 2 min
2s O
O O
O H
H
O
O
O
H
H
O
O
O
H
H
H
H
H O
O O
O
O
H
H
O
O
O
O
O O
O
O
O O
H
H
O
O
H
H
H
H
O
O O
O
H O
O
O
H
H
O
H O
O
O
H
H
O
O
O
O H
H O
O O
O O
H H
H
O
O O O
H
O
O
O H H
O
O
O
O O
O
OH
O
O O
O
O
O O
O
O
O
H
O H
H
H
Figure 1. Schematic illustrating 9AC rod bending. The darkened oval indicates the region consisting of 9AC photodimer created by 2PE.
2. Results
femtosecond lasers was used. Figure 2 shows the one-photon absorption spectra of 9AC in a solid film. The solid-phase spectrum is red-shifted due to its polar environment but still retains the characteristic vibronic peaks of the anthracene monomer absorption. Also shown in Figure 2 is the 2PE spectrum for the 9AC nanorod fluorescence. The 2PE spectrum follows the one-photon absorption of the solid in the wavelength range 380–450 nm, then increases rapidly at shorter wavelengths. This behavior of the 2PE spectrum has been observed for anthracene and its derivatives.[21–23] Vibronic coupling effects and environmental perturbations can make the lowest energy electronic state accessible by both one- and twophoton absorption events.[24] The increase in the 2PE spectrum
1.0
0.8
Absorbance
In this paper, we examine whether it is possible to control the magnitude, rate, and orientation of the photoinduced deformations in the nanorods. In order to precisely control the location and magnitude of the bend, two-photon excitation (2PE) is used to localize the region of reacted molecules in 3D space. 2PE allows us to take advantage of the superior penetration characteristics of the near-IR as opposed to the UV light used in our previous work. Using this method, we demonstrate the controlled bending of both 200- and 35-nm-diameter nanorods, which are the smallest photomechanical structures yet demonstrated experimentally. We also show that sequential bending of longer segments can, in some cases, lead to actual translation of entire rods. We hypothesize that the bending mechanism results from the lattice mismatch between monomer and dimer crystals within a photoexposed nanorod. By measuring the bend angles as a function of irradiation time and intensity, we show that both the rate and magnitude of deformation are consistent with a simple kinetic model where the bend angle is linearly proportional to the fraction of dimerized molecules. This quantitative understanding provides a foundation for both engineering the nanoscale photomechanical response, and for a more detailed chemical understanding of the phenomenon. Taken together, the results in this paper indicate that molecular crystal nanostructures have significant promise as photoactuators and should be examined in more detail.
0.6
0.4
0.2
0.0 350
2.1. Characterization of the Two-Photon Response of 9AC Nanorods In order to generate a region of high excitation density localized in 3D, multiphoton excitation via highly focused small 2009, 5, No. 24, 2902–2909
400
450
Wavelength (nm) Figure 2. One-photon absorbance (solid line) and two-photon excitation spectra (squares) of solid 9AC. The two-photon profile has been shifted to half the excitation wavelength to compare the two-photon and onephoton absorption profiles.
ß 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
2903
full papers
C. J. Bardeen et al.
at shorter wavelengths can be attributed to absorption to the higher-lying states that are one-photon forbidden but twophoton allowed.[22] These spectra show that 9AC nanorods preserve the same general two-photon absorption behavior of anthracene derivatives and are thus good candidates for multiphoton excitation.
2.2. Dispersion of 9AC Nanorods via Surfactant Treatment Given that the 9AC nanorods can undergo 2PE using nearIR femtosecond pulses, the next step is to demonstrate that 2PE can induce reversible bending similar to that observed using one-photon excitation. To accomplish bending in a routine manner for our 9AC nanorods, we modified our preparation conditions to obtain a high yield of isolated, active rods. Earlier we used a different method, silica coating, to better disperse the 9AC nanorods in aqueous solution.[25] Those core–shell nanorods still exhibited bending behavior under UV illumination, but at a much reduced rate. Under the assumption that the silica coating inhibited the bending, we decided to use a softer surfactant coating to produce dispersed nanorods. By pretreating the alumina templates with an acidic sodium dodecyl sulfate (SDS) solution,[26,27] we could form each rod inside a shell of SDS, resulting in much better dispersion in aqueous solution. As in our previous work on molecular crystal nanorods, transmission electron microscopy (TEM) images showed highly oriented crystal domains that extended over at least several micrometers within the rods. Furthermore, the SDStreated nanorods showed much more consistent bending behavior than the uncoated rods of our earlier study. At least 60% of the SDS-coated rods showed bending behavior, as compared to 10% of the uncoated rods. Since TEM images showed that both the treated and untreated rods were both highly crystalline, we do not think the higher yield of bending rods is due to differences in crystallinity. Rather, we think it is due to differences in aggregation and surface adhesion. Rods that are attached either to each other or to a surface tend to be more difficult to bend, since the adhesion competes with the bending forces.
magnitude of the bend for a given rod were highly reproducible. After the initial bend, the size of the bend was reproducible after at least 5 cycles. The decrease in bend magnitude could be due to a number of factors: photobleaching of vulnerable molecules in the rod, changes in the rod–surface interaction or rod orientation after the first bending event, or changes in the crystallinity after the initial bend. In most cases, the nanorod remained in place while one segment was displaced by the bend, as shown in Figure 3. Since the ‘‘scallop theorem’’ says that a simple bend cannot result in a net translation for a free rod in a low Reynolds number environment,[28] we attempted to use more complicated illumination conditions in an attempt to induce net translation of the rod. By illuminating longer rod segments, a monomer–dimer interface resulted in the asymmetric bending of an entire segment, rather than a symmetric bend around a single point. Using this approach, we were sometimes able to induce a type of walking motion in a nanorod, an example of which is shown in Figure 4. In this case, the rod–surface interactions were apparently weak enough to allow the entire rod to be translated by bending of entire segments, rather than by single-point bending. It is likely that sequential detachment and reattachment of different segments to the surface play a role in this type of motion, since we could typically only take a few ‘‘steps’’ before it became impossible to induce further motion in the rods where this was observed.
2.3.2. Bending of 35-nm-Diameter Rods The 35-nm-diameter nanorods exhibited behavior similar to that of the 200-nm-diameter rods. These smaller-diameter rods presented more experimental challenges since they have lower contrast in the phase microscope and also require the use of different-diameter templates so that lower numbers are produced. Nevertheless, we were able to locate and bend individual 35-nm rods, which showed roughly the same kinetics
2.3. Control of Reversible Bending in 9AC Nanorods 2.3.1. Bending of 200-nm-Diameter Rods For non-aggregated rods, it was straightforward to observe reversible bending behavior. Figure 3 shows a typical sequence of optical microscopy images of a single 200-nm-diameter 9AC nanorod reversibly bending and recovering after excitation at different locations along the rod. Illumination of a 1.5-mm-diameter spot along the nanorod induced a reproducible bend centered at that location. The direction and
2904 www.small-journal.com
Figure 3. Single 9AC rod 200 nm in diameter exposed to 800-nm light (focused into a 1.5-mm spot) in phosphoric acid solution with surfactant. The rod in (a) is irradiated for 2 s in a spot near the top (white circle indicates location and diameter of laser spot), resulting in a bend near the top of the rod in (b). After 2 min in darkness it relaxes to its former configuration (c). This cycle is then continued with irradiation in one spot (d,f), where the bend is indicated by the white circle, followed by relaxation (e,g). Scale bar ¼ 15 mm.
ß 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2009, 5, No. 24, 2902–2909
Bending Motions in Molecular Crystal Nanorods
Figure 4. Single 9AC nanorod moving in stepwise fashion as a result of repeated irradiation. a) The top portion of the rod, enclosed by the white ellipse, has been irradiated and pivots to the left (b), where the outlined ghost image shows the position of the rod from (a). The top portion of the rod is irradiated again (c) and as it relaxes, the top of the rod pivots towards the left again (d). e) The bottom portion of the rod is irradiated resulting in the bottom of the rod pivoting towards the left in (f). The stationary rod at left serves as a reference point for the motion. Scale bar ¼ 10 mm.
2.3.3. Quantification of Magnitude and Rate of Rod Bending Figures 3–5 show how 2PE can be used to induce bending dynamics in single 9AC nanorods. We now turn to an analysis of quantities like angles and rates of rod bending under illumination (kbend) and recovery in the dark (kunbend). All the rods exhibited exponential rises followed by a constant plateau. Figure 6a shows a plot of bend angle versus illumination time averaged over a population of rods from the same preparation batch. This data yields an average maximum bend angle of 248 and a bend rate, kbend ¼ 0.81 s1. There is a large distribution of rates in the bending data, however, and when individual curves are analyzed we find that these times can vary by a factor of 10 (0.3 s1 to 3.2 s1). In other words, the averaged data in Figure 6a provides only an approximate guide to the bending rate for an individual rod. Once the light was turned off, the rods straightened out again,
a 25
20
Bend Angle (deg)
as the 200-nm-diameter rods. Figure 5 shows a reversible bend induced in a single 35-nm rod under the same exposure conditions as the 200-nm-diameter rods.
15
10
5
0 0
1
2
3
4
5
6
7
8
Irradiation Time (s)
b 25
Bend Angle (deg)
20
15
10
5
0 0
100
200
300
400
500
Recovery Time (s)
Figure 5. Single 9AC rod 35 nm in diameter exposed to 800-nm light under the same conditions used for bending the 200-nm rods. In the left frame, the rod is at rest. The middle frame shows the rod after a 2-s illumination of the region enclosed by the black circle, resulting in a 178 bend. After 2 min in darkness,the rod revertsback to its originalshape in the right frame. Scale bar ¼ 5 mm. small 2009, 5, No. 24, 2902–2909
Figure 6. a) Average bending curve for 9AC rods excited at constant intensity in a single exposure lasting several seconds. This curve was generated by averaging data points acquired from bending runs on a sample of template-treated 9AC rods. Bending runs that terminated beforetheotherrunswereextrapolated asa flatlineatthefinalbend angle for each rod. b) The average recovery curve of the rods begins after irradiation ceases. This curve was generated by averaging data points acquired from runs of bending on a similar sample of rods. Fit lines shown are generated using Equations (3) and (4) for the bend and recovery curves, respectively.
ß 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
2905
full papers
C. J. Bardeen et al.
and the kunbend values showed much less variation. Figure 6b shows the decay curve averaged over different rods, along with the exponential fit with kunbend ¼ 0.0090 s1. In this case, the kunbend rates for individual rods only ranged from 0.0084 s1 to 0.0095 s1, or a factor of 1.13. Furthermore, the unbending followed the same decay law in the dark as the fluorescence recovery, indicating that it is molecular in origin. To explain the large variation in kbend rates, we consider three possible origins. The first is that different rods have different chemical characteristics (crystallinity, defect density, impurity levels) that determine the bending rate. Inspection of rod samples under TEM failed to reveal observable variations in crystallinity. Furthermore, morphology effects that lead to variations in kbend would be expected to affect kunbend as well. Since the distribution of kunbend values was much narrower than that of the kbend values, there is no independent evidence for variations in rod structure that would affect the deformation rates. A second possibility is that variations in laser intensity cause different excitation rates and thus different bending rates. The rods were confined to a 1-mm-thick well for observation, which is much shorter than the Rayleigh range of the focused laser beam. For each rod, the focus was optimized manually within the sample before staring the bending measurements. While the optimum focusing position could be determined, the quality of the focus may have varied due to imperfect focusing, local variations in scattering, or the proximity of dielectric interfaces. The nonlinear dependence of kbend on intensity described below in Equation (5a) would enhance the effects of small intensity variations. The last explanation for the observed variation in kbend is the presence of different amounts of adhesion to the glass surface by different rods. The evidence for surface effects includes the observation that uncoated rods exhibited bending behavior much less frequently than the SDScoated rods, and that individual rods were observed to bend more rapidly during the second and subsequent bending sequences. Taken together, these observations suggest that the variations in kbend originate from the last two factors, which are due to experimental nonidealities instead of variations in rod structure. The fastest observed kbend values probably provide a lower bound for the intrinsic kbend value of free rods.
3. Discussion 3.1. Molecular Origins of Reversible Bending Our earlier work on 9-tertbutyl-anthroate (9TBAE) showed that nanorods composed of this compound could undergo a nonreversible photodimerization that involved a crystal-to-crystal transition.[15] It is reasonable to assume that the 9AC photodimerization also proceeds in a crystal-to-crystal fashion. The bending is most likely due to the interface between reacted and unreacted 9AC crystal regions in the rods. This reasoning is supported by the observation that illumination of an entire segment results in asymmetric bending while illumination at a single point gives rise to a symmetric bend. Point illumination gives rise to two interfaces between reacted and unreacted molecules, resulting in two bends around the central point. Illumination of a segment from one end to a middle point, as shown in Figure 4, leaves only one interface
2906 www.small-journal.com
where the bend can occur. The bending then appears to be asymmetric since only one part of the rod moves while the other remains stationary. The physical origin of the bending could be strain arising from the lattice mismatch between unreacted and partially reacted regions in the rod. If the strain in any one region along the rod is proportional to the fraction of reacted molecules in that region,[29] then the bend angle should also be linearly proportional to the fraction of reacted molecules. A somewhat related mechanism posits that the crystal lattices of the monomer and dimer are sufficiently similar that the bend angle can vary smoothly with concentration of dimers within a specific region. In this case, the bend would be due to a structural effect, rather than to strain. In either case, we assume that the bend angle can vary continuously with the fraction of photoinduced dimers. If this assumption holds, then it is straightforward to generate a simple kinetic model for the lightinduced bending. For 2PE, the monomer and dimer populations Nmon and Ndim are governed by the following two differential equations: @Nmon ¼ 2dI 2 ’dim Nmon þ 2kdissoc Ndim @t
(1a)
@Ndim ¼ þdI 2 ’dim Nmon kdissoc Ndim @t
(1b)
where d is the two-photon absorption cross section, I is the light intensity, fdim is the quantum yield of photodimerization, and kdissoc is the rate of dimer dissociation back into monomers. The kinetics of reactive molecules in 1D chains (the 1D stacks of 9AC molecules) are more complicated than reflected in Equations (1a) and (1b) but can be approximated by assuming that a fraction P of the monomers remains after completion of the reaction. The existence of P is due to the fact that in a chain of reactive molecules, each molecule can react with either its left- or right-hand neighbor. As the reaction proceeds, some molecules find that both neighbors have already reacted, leaving the central molecule without a partner. In the ideal case of equal reaction probabilities between all molecules, it has been shown that P ¼ 0.13.[30,31] Note that conservation of the total number of molecules 0 means that Nmon þ 2Ndim ¼ Nmon and Equations (1a) and (1b) are not independent. This means that the kinetics can be completely described in terms of a single variable, which we define as X¼
2Ndim ¼ fraction of converted monomers 0 Nmon
(2)
As described above, the bend angle is proportional to the induced strain, which is proportional to X. a is defined to be the proportionality constant. During illumination, we neglect kdissoc and find ubend ¼ aX ¼ að1 PÞ 1 exp 2dI 2 ’dim t
(3)
If the illumination is then turned off, I ¼ 0 and we can no longer neglect the kdissoc term, resulting in a new time-
ß 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2009, 5, No. 24, 2902–2909
Bending Motions in Molecular Crystal Nanorods
dependence for the bend angle: ubend ¼ aX ¼ aX ðt ¼ 0Þ exp½kdissoc t
(4)
Equations (3) and (4) predict an exponential increase in bend angle under illumination followed by an exponential decay after the excitation light is turned off. We observe both phenomena experimentally. Equations (3) and (4) also predict values for the experimentally observed kbend and kunbend rates: kbend ¼ 2dI 2 ’dim
(5a)
kunbend ¼ kdissoc
(5b)
The value for kunbend is similar to the recovery time of the rod fluorescence after photoreaction, which should reflect kdissoc since dissociation results in a recovery of the strong monomer fluorescence. The value for kbend involves more variables but we can make an order of magnitude estimation. If we take a rough value of d ¼ 1051 cm4 s1 for crystalline anthracene[22] and estimate fdim ¼ 0.1, and combine them with our experimental value of I ¼ 4.2 1028 photons per cm2 s1 (assuming a 150-fs pulsewidth and a 1.5-mm-diameter spot), using Equation (5a) we find kbend 1 s1, which is close to the experimental value of 0.81 s1. Note that for the case of 2PE in a 200-nm-diameter nanorod, attenuation of the exciting beam as it propagates through the rod should be negligible, since the characteristic length for two-photon absorption is estimated to be greater than 1 cm due to the low d value. The calculated kbend value could change by at least an order of magnitude in either direction due to uncertainties in d, fdim, and I, but serves to demonstrate that for reasonable values of these parameters, we can calculate a value for kbend in good agreement with the experimental value. This indicates that Equations (5a) and (5b) can be used as a starting point to analyze the bending dynamics.
3.2. Limitations on Control of Nanorod Bending The results presented above demonstrate that it is possible to control both the location (by laser spot placement) and magnitude (by laser intensity and/or exposure time) of the induced bending. One aspect we were unable to control was the direction of the bend. Attempts to vary the bend direction by adjusting the position of the spot with respect to the rod axis were unsuccessful; a given rod always bent in the same direction, no matter what illumination conditions were used. The bend direction was consistent along a given rod as well, as can be seen from Figure 3. This observation suggests that the bend direction is determined by an intrinsic property of the nanorod, as opposed to an environmental factor like surface adhesion or irradiation profile. While defects may play a role, it is unlikely that defects would induce the same behavior at different points along a 60-mm-long rod. A more reasonable explanation involves the anisotropy of the 9AC crystal in the nanorod. Powder X-ray diffraction and TEM measurements have confirmed the preferential orientation of the 9AC crystal along the long axis of the nanorod.[25] However, knowing the orientation of the crystal with respect to the rod’s long axis still small 2009, 5, No. 24, 2902–2909
leaves an ambiguity as to the axial rotation of the crystal within the cylindrical rod. For example, if the c-axis of the crystal corresponds to the long axis of the rod, the angles of the a- and baxes within a specific rod lying on the surface are unknown. In general, these axes will be randomly oriented. In our simple example, if the preferred bend direction is along the a-axis, then the bend direction within a given rod will always be the same (along the a-axis) but will be different for different rods, since some will lie with the a-axis pointing to the left while others lie with the a-axis pointing to the right. The random orientation of the bend direction on the surface may also explain the observed variability in bend magnitude, since if the bend direction is not oriented parallel to the surface, the amount of observable bend angle in the plane of observation will decrease. Finally, it should be noted that although numerous methods have been reported for aligning nanowires with respect to their long axis, we were unable to find experiments where the rotational angle around its long axis can be controlled. If the intrinsic nanorod crystal orientation determines the bend direction, this would be an important difference with respect to liquid-crystal elastomers doped with photochromic molecules. In those types of structures, the bend direction is determined by a combination of extrinsic factors, like asymmetry in the light absorption, as well as by intrinsic properties like liquid crystal alignment.[29,32–34] For example, in azobenzene-doped elastomers, changing the polarization and intensity of light affords some control over the deformation but require that the light intensity varies within the structure. This intensity variation is usually due to Beer’s Law attenuation of the exciting beam, but this attenuation becomes negligible when nanometer-scale structures are used. For molecular crystal nanorods, on the other hand, there would not be a size limit for directional bending, as long as good crystallinity is maintained.
4. Conclusion We have expanded our earlier work on the reversible photoinduced bending of 9AC molecular crystal nanorods. We have shown that highly localized 2PE can induce bending at specific locations along a single rod, and that the phenomenon can be scaled from 200- to 35-nm-diameter rods. Furthermore, we have developed a simple kinetic model that can quantitatively describe the bending and relaxation behavior of individual rods. Remaining challenges include controlling the axial orientation of the rods in order to control the bending direction, and better control of rod–surface interactions to obtain more reproducible bend magnitudes and rates. Despite these challenges, this work shows that photomechanical effects in molecular crystal nanorods can be controlled in a reproducible and precise manner, and thus these materials have promise for applications in the area of nanomanipulation.
5. Experimental Section Preparation of 9AC nanorods: The 9AC nanorods were prepared using a template synthesis method that employs
ß 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
2907
full papers
C. J. Bardeen et al.
commercially available anodized alumina filters (Whatman Anodisc-13, 200-nm pore diameter, membrane thickness 60 mm, pore density 109 cm2).[16,35] To aid in the isolation of the nanorods, the templates were pretreated with a surfactant (SDS). A drop of SDS solution (1 wt%), acidified to pH 1.8 by addition of sulfuric acid, was deposited on both sides of the template and allowed to dry for 18 h. Then a solution of 13 mg 9AC (Aldrich, 99% purity), in 0.2 mL tetrahydrofuran (THF; HPLC grade 99.9% purity) was slowly deposited on both sides of the Anodisc. The template was then placed on a plastic stand in a glass jar containing filter paper soaked with additional THF (2 mL). The jar was sealed except for two holes poked in the lid of the jar that allowed air to flow at a slow rate through a gas bubbler (about 1 bubble per 3 s) to slowly evaporate the THF. After standing at room temperature for 36 h, the template was removed and polished with 1500 grit sandpaper to remove excess 9AC. The template was then dissolved in a solution of phosphoric acid (50% aqueous) creating a suspension of 9AC rods. To 2 mL of this suspension, an additional 0.4 mL SDS solution (1 wt% aqueous) was added to produce the final rod solution used for bending experiments. To make 35-nm rods, different alumina templates (Synkera OA-35-50-13 AAO, 35-nm pore diameter, membrane thickness 50 mm, pore density 1010 cm2) were employed and 9AC (6 mg in 0.1 mL THF) was deposited on both sides of the template. Observation of nanorod bending: To observe nanorod bending, 1-mm-deep well plates were made by coating microscope cover slips with a layer of gold 1-mm in thickness with a 1 cm2 square masked off in the middle. A drop of nanorod solution was deposited in the well and sandwiched by placing another cover slip on top. An inverted optical microscope equipped with phase contrast rings (Olympus IX70) and a 60 (1.2 NA) oil immersion objective was used for imaging the sample. To excite the rods, 150 fs 800-nm pulses generated by a Ti:sapphire laser (Spectra Physics-Mai Tai) at a 82-MHz repletion rate were focused onto the sample from below, resulting in a spot 1.5 mm in diameter (full width at half maximum). The peak intensity during each femtosecond pulse is 1010 W cm2, typical of intensities used for two-photon fluorescence imaging and well below the damage threshold of the crystal. For typical runs, bending was achieved by exposing a segment of the rod to the laser at a power of 100 mW for several seconds. To monitor the time resolved bending of the rods, rods were continuously monitored (at a rate of 30 frames per second) during and after laser exposure. Measurement of two-photon absorbtion of nanorods: To measure the 2PE spectrum of the 9AC nanorods, an aqueous rod suspension was placed on a microscope cover slip and imaged in the microscope with a 20 objective. A photomultiplier tube (Hamamatsu R5600U) was coupled to the camera port on the microscope to collect the fluorescence signal produced by exciting a bundle of nanorods at different wavelengths across the tunable range of the Mai-Tai laser (710–920 nm, 10-nm steps). A chopper (Stanford Research Systems SR 540) coupled with a Lock-in amplifier (SR 830 DSP) were used to detect the fluorescence signal, which was proportional to the intensity squared.
2908 www.small-journal.com
Acknowledgements Discussions with Prof. Rabih O. Al-Kaysi are gratefully acknowledged. We thank Dexter Humphrey and the Center for Nanoscale Science and Engineering (CNSE) at UCR for the use of the electron-beam evaporator. Electron microscopy measurements on the nanorods were performed at the Central Facility for Advanced Microscopy and Microanalysis (CFAMM) at UCR. This work was funded by the National Science Foundation, grant DMR-0907310.
[1] T. Ikeda, J. Mamiya, Y. Yu, Angew. Chem. Int. Ed. 2007, 46, 506– 528. [2] H. Finkelmann, E. Nishikawa, G. G. Pereira, M. Warner, Phys. Rev. Lett. 2001, 87, 015501–015504. [3] M. Yamada, M. Kondo, R. Myasato, Y. Naka, J. Mamiya, M. Kinoshita, A. Shishido, Y. Yu, C. J. Barrett, T. Ikeda, J. Mater. Chem. 2009, 19, 60–62. [4] Y. Yu, M. Nakano, T. Ikeda, Nature 2003, 425, 145. [5] M. Camacho-Lopez, H. Finkelmann, P. Palffy-Muhoray, M. Shelley, Nat. Mater. 2004, 3, 307–310. [6] M. Yamada, M. Kondo, J. Mamiya, Y. Yu, M. Kinoshita, C. J. Barrett, T. Ikeda, Angew. Chem. Int. Ed. 2008, 47, 4986–4988. [7] T. J. White, N. V. Tabiryan, S. V. Serak, U. A. Hrozhyk, V. P. Tondiglia, H. Koerner, R. A. Vaia, T. J. Bunning, Soft. Mater. 2008, 4, 1796– 1798. [8] K. D. Harris, R. Cuypers, P. Scheibe, C. L. v. Oosten, C. W. M. Bastiaansen, J. Lub, D. J. Broer, J. Mater. Chem. 2005, 15, 5043–5048. [9] M. Irie, S. Kobatake, M. Horichi, Science 2001, 291, 1769–1772. [10] I. Colombier, S. Spagnoli, A. Corval, P. L. Baldeck, M. Giraud, A. Leaustic, P. Yu, M. Irie, J. Chem. Phys. 2007, 126, 011101/1– 011101/3. [11] S. Kobatake, S. Takami, H. Muto, T. Ishikawa, M. Irie, Nature 2007, 446, 778–781. [12] K. Uchida, S. Sukata, Y. Matsuzawa, M. Akazawa, J. J. D. d. Jong, N. Katsonis, Y. Kojima, S. Nakamura, J. Areephong, A. Meetsma, B. L. Feringa, Chem. Commun. 2008, 326–328. [13] D. J. Flannigan, V. A. Lobastov, A. H. Zewail, Angew. Chem. Int. Ed. 2007, 46, 9206–9210. [14] D. J. Flannigan, P. C. Samartzis, A. Yurtsever, A. H. Zewail, Nano Lett. 2009, 9, 875–881. [15] R. O. Al-Kaysi, A. M. Muller, C. J. Bardeen, J. Am. Chem. Soc. 2006, 128, 15938–15939. [16] R. O. Al-Kaysi, C. J. Bardeen, Adv. Mater. 2007, 19, 1276–1280. [17] R. O. Al-Kaysi, R. J. Dillon, J. M. Kaiser, L. J. Mueller, G. Guirado, C. J. Bardeen, Macromolecules 2007, 40, 9040–9044. [18] D. K. Bucar, L. R. MacGillivray, J. Am. Chem. Soc. 2007, 129, 32–33. [19] S. Takahashi, H. Miura, H. Kasai, S. Okada, H. Oikawa, H. Nakanishi, J. Am. Chem. Soc. 2002, 124, 10944–10945. [20] Y. Ito, H. Fujita, J. Org. Chem. 1996, 61, 5677–5680. [21] J. Wolf, G. Hohlneicher, Chem. Phys. 1994, 181, 185–208. [22] I. Webman, J. Jortner, J. Chem. Phys. 1969, 50, 2706–2716. [23] W. J. Yang, M. S. Seo, X. Q. Wang, S. J. Jeon, B. R. Cho, J. Fluoresc. 2008, 18, 403–411. [24] A. Bree, R. Zwarich, Chem. Phys. 1993, 170, 185–192. [25] R. O. Al-Kaysi, R. J. Dillon, L. Zhu, C. J. Bardeen, J. Colloid Interface Sci. 2008, 327, 102–107. [26] T. Saitoh, K. Taguchi, M. Hiraide, Anal. Chim. Acta 2002, 454, 203– 208.
ß 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2009, 5, No. 24, 2902–2909
Bending Motions in Molecular Crystal Nanorods
[27] C. Pan, L. Zhang, J. Zhu, J. Luo, Z. Cheng, C. Wang, Nanotechnology 2007, 18, 015302/1–015302/6. [28] E. M. Purcell, Am. J. Phys. 1977, 45, 3–11. [29] M. L. Dunn, J. Appl. Phys. 2007, 102, 013506/1–013506/7. [30] E. R. Cohen, H. Reiss, J. Chem. Phys. 1963, 38, 680–691. [31] J. Even, M. Bertault, J. Chem. Phys. 1999, 110, 1087–1096. [32] M. Warner, L. Mahadevan, Phys. Rev. Lett. 2004, 92, 134302/1–134302/4.
small 2009, 5, No. 24, 2902–2909
[33] D. Corbett, C. L. v. Oosten, M. Warner, Phys. Rev. A 2008, 78, 013823/1–013823/4. [34] C. L. v. Oosten, D. Corbett, D. Davies, M. Warner, C. W. M. Bastiaansen, D. J. Broer, Macromolecules 2008, 41, 8592–8596. [35] R. O. Al-Kaysi, C. J. Bardeen, Chem. Commun. 2006, 1224–1226.
ß 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Received: May 26, 2009 Published online: October 23, 2009
www.small-journal.com
2909
Using Two-Photon Excitation to Control Bending Motions in Molecular-Crystal Nanorods Jacob T. Good, Jonathan J. Burdett, and Christopher J. Bardeen*
Molecular-crystal nanorods composed of 9-anthracenecarboxylic acid can undergo reversible bending due to molecular-level geometry changes associated with the photodimerization of the molecules in the crystal lattice. The use of highly focused near-IR femtosecond laser pulses results in twophoton excitation of micrometer-scale regions and is used to induce transient bends at various locations along the length of a single 200-nm-diameter nanorod. Bending can be observed in nanorods with diameters as small as 35 nm, and translational motion of a single nanorod could be induced by sequential bending of longer segments. A kinetic model is presented that quantitatively describes the bending and relaxation dynamics of individual rods. The results of this work show that it is possible to use laser excitation conditions to control the location, rate, and magnitude of photodeformations in these nanorods. The ability to control the motion of these ultrasmall photomechanical structures may be useful for manipulating objects on the nanoscale.
1. Introduction The manipulation of nano- and micrometer scale objects is a central goal in many areas of science, from physics to biology. Photons are ideal for controlling motion since they can penetrate deep into 3D samples like biological media and do not require physical contact between the controller and object. To convert the photons into mechanical work, small-scale photomechanical actuators are required. The development of photomechanical actuators has centered on the use of photochemically active chromophores, like azobenzene and spiropyran, embedded in polymers and liquid-crystal elastomers.[1,2] Progress in this field has been rapid, with recent demonstrations of artificial inchworms, swimmers, motors, and oscillators.[3–8] Recently, a new type of organic photomechanical material, based on single-component photochemically active molecular crystals, has been demonstrated. Diarylethene
[] Prof. C. J. Bardeen, J. T. Good, J. J. Burdett Department of Chemistry University of California at Riverside Riverside, CA 92521 (USA) E-mail: [email protected] DOI: 10.1002/smll.200900895
2902
Keywords:
bending motions nanorods photomechanical control two-photon excitation
microcrystals and thin crystal plates have been shown to undergo large deformations driven by intramolecular ringopening and closing reactions activated by photons of different wavelengths,[9–12] while Cu-TCNQ (TCNQ ¼ 7,7,8,8-tetracyanoquinodimethane) crystals have also shown reversible deformations due to photoinduced charge transfer.[13,14] Our group has investigated the behavior of molecular crystalline nanorods composed of anthracene derivatives that undergo a 4þ4 photodimerization reaction. In this type of reaction, a single anthracene absorbs a photon and reacts with a neighboring molecule to form a bridged dimer, as illustrated in Figure 1.[15–17] The resulting photodimer is typically stable unless dissociation is induced by a UV photon or heat. The nanoscale dimensions of rods composed of these anthracene derivatives permit them to undergo crystal-to-crystal photoreactions that would cause a macroscopic crystal to disintegrate.[18,19] In the case of 9-anthracene carboxylic acid (9AC), the photodimer spontaneously dissociates within a few minutes, resetting the system.[20] The initial demonstrations of reversible photomechanical molecular crystals provide evidence that they could be useful as small-scale photoactuators. However, to assess whether the nanorod photoresponse can be harnessed to generate controlled, microscale displacements, several aspects of their photomechanical response must be understood in more detail.
ß 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2009, 5, No. 24, 2902–2909
Bending Motions in Molecular Crystal Nanorods
H
H
H
O
O
H H
H O
O
O
O
H O
O
O
O H
O
O
H
H
H
O H
O
O O
O
H O
O
O
O
O
O
O
H
H
H O
O H O
O
O
H
H
H
H
H
OH O
O
O
O O
O
O
H O
O
O
O
H
H
H
O
O
O
H O
O
O
O H
O
O
O
O
H
O
O
O
O
O
O
H
H O
O O
O
O
O
O H
O H
O
H
O
H
O
O
O
O
kT 2 min
2s O
O O
O H
H
O
O
O
H
H
O
O
O
H
H
H
H
H O
O O
O
O
H
H
O
O
O
O
O O
O
O
O O
H
H
O
O
H
H
H
H
O
O O
O
H O
O
O
H
H
O
H O
O
O
H
H
O
O
O
O H
H O
O O
O O
H H
H
O
O O O
H
O
O
O H H
O
O
O
O O
O
OH
O
O O
O
O
O O
O
O
O
H
O H
H
H
Figure 1. Schematic illustrating 9AC rod bending. The darkened oval indicates the region consisting of 9AC photodimer created by 2PE.
2. Results
femtosecond lasers was used. Figure 2 shows the one-photon absorption spectra of 9AC in a solid film. The solid-phase spectrum is red-shifted due to its polar environment but still retains the characteristic vibronic peaks of the anthracene monomer absorption. Also shown in Figure 2 is the 2PE spectrum for the 9AC nanorod fluorescence. The 2PE spectrum follows the one-photon absorption of the solid in the wavelength range 380–450 nm, then increases rapidly at shorter wavelengths. This behavior of the 2PE spectrum has been observed for anthracene and its derivatives.[21–23] Vibronic coupling effects and environmental perturbations can make the lowest energy electronic state accessible by both one- and twophoton absorption events.[24] The increase in the 2PE spectrum
1.0
0.8
Absorbance
In this paper, we examine whether it is possible to control the magnitude, rate, and orientation of the photoinduced deformations in the nanorods. In order to precisely control the location and magnitude of the bend, two-photon excitation (2PE) is used to localize the region of reacted molecules in 3D space. 2PE allows us to take advantage of the superior penetration characteristics of the near-IR as opposed to the UV light used in our previous work. Using this method, we demonstrate the controlled bending of both 200- and 35-nm-diameter nanorods, which are the smallest photomechanical structures yet demonstrated experimentally. We also show that sequential bending of longer segments can, in some cases, lead to actual translation of entire rods. We hypothesize that the bending mechanism results from the lattice mismatch between monomer and dimer crystals within a photoexposed nanorod. By measuring the bend angles as a function of irradiation time and intensity, we show that both the rate and magnitude of deformation are consistent with a simple kinetic model where the bend angle is linearly proportional to the fraction of dimerized molecules. This quantitative understanding provides a foundation for both engineering the nanoscale photomechanical response, and for a more detailed chemical understanding of the phenomenon. Taken together, the results in this paper indicate that molecular crystal nanostructures have significant promise as photoactuators and should be examined in more detail.
0.6
0.4
0.2
0.0 350
2.1. Characterization of the Two-Photon Response of 9AC Nanorods In order to generate a region of high excitation density localized in 3D, multiphoton excitation via highly focused small 2009, 5, No. 24, 2902–2909
400
450
Wavelength (nm) Figure 2. One-photon absorbance (solid line) and two-photon excitation spectra (squares) of solid 9AC. The two-photon profile has been shifted to half the excitation wavelength to compare the two-photon and onephoton absorption profiles.
ß 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
2903
full papers
C. J. Bardeen et al.
at shorter wavelengths can be attributed to absorption to the higher-lying states that are one-photon forbidden but twophoton allowed.[22] These spectra show that 9AC nanorods preserve the same general two-photon absorption behavior of anthracene derivatives and are thus good candidates for multiphoton excitation.
2.2. Dispersion of 9AC Nanorods via Surfactant Treatment Given that the 9AC nanorods can undergo 2PE using nearIR femtosecond pulses, the next step is to demonstrate that 2PE can induce reversible bending similar to that observed using one-photon excitation. To accomplish bending in a routine manner for our 9AC nanorods, we modified our preparation conditions to obtain a high yield of isolated, active rods. Earlier we used a different method, silica coating, to better disperse the 9AC nanorods in aqueous solution.[25] Those core–shell nanorods still exhibited bending behavior under UV illumination, but at a much reduced rate. Under the assumption that the silica coating inhibited the bending, we decided to use a softer surfactant coating to produce dispersed nanorods. By pretreating the alumina templates with an acidic sodium dodecyl sulfate (SDS) solution,[26,27] we could form each rod inside a shell of SDS, resulting in much better dispersion in aqueous solution. As in our previous work on molecular crystal nanorods, transmission electron microscopy (TEM) images showed highly oriented crystal domains that extended over at least several micrometers within the rods. Furthermore, the SDStreated nanorods showed much more consistent bending behavior than the uncoated rods of our earlier study. At least 60% of the SDS-coated rods showed bending behavior, as compared to 10% of the uncoated rods. Since TEM images showed that both the treated and untreated rods were both highly crystalline, we do not think the higher yield of bending rods is due to differences in crystallinity. Rather, we think it is due to differences in aggregation and surface adhesion. Rods that are attached either to each other or to a surface tend to be more difficult to bend, since the adhesion competes with the bending forces.
magnitude of the bend for a given rod were highly reproducible. After the initial bend, the size of the bend was reproducible after at least 5 cycles. The decrease in bend magnitude could be due to a number of factors: photobleaching of vulnerable molecules in the rod, changes in the rod–surface interaction or rod orientation after the first bending event, or changes in the crystallinity after the initial bend. In most cases, the nanorod remained in place while one segment was displaced by the bend, as shown in Figure 3. Since the ‘‘scallop theorem’’ says that a simple bend cannot result in a net translation for a free rod in a low Reynolds number environment,[28] we attempted to use more complicated illumination conditions in an attempt to induce net translation of the rod. By illuminating longer rod segments, a monomer–dimer interface resulted in the asymmetric bending of an entire segment, rather than a symmetric bend around a single point. Using this approach, we were sometimes able to induce a type of walking motion in a nanorod, an example of which is shown in Figure 4. In this case, the rod–surface interactions were apparently weak enough to allow the entire rod to be translated by bending of entire segments, rather than by single-point bending. It is likely that sequential detachment and reattachment of different segments to the surface play a role in this type of motion, since we could typically only take a few ‘‘steps’’ before it became impossible to induce further motion in the rods where this was observed.
2.3.2. Bending of 35-nm-Diameter Rods The 35-nm-diameter nanorods exhibited behavior similar to that of the 200-nm-diameter rods. These smaller-diameter rods presented more experimental challenges since they have lower contrast in the phase microscope and also require the use of different-diameter templates so that lower numbers are produced. Nevertheless, we were able to locate and bend individual 35-nm rods, which showed roughly the same kinetics
2.3. Control of Reversible Bending in 9AC Nanorods 2.3.1. Bending of 200-nm-Diameter Rods For non-aggregated rods, it was straightforward to observe reversible bending behavior. Figure 3 shows a typical sequence of optical microscopy images of a single 200-nm-diameter 9AC nanorod reversibly bending and recovering after excitation at different locations along the rod. Illumination of a 1.5-mm-diameter spot along the nanorod induced a reproducible bend centered at that location. The direction and
2904 www.small-journal.com
Figure 3. Single 9AC rod 200 nm in diameter exposed to 800-nm light (focused into a 1.5-mm spot) in phosphoric acid solution with surfactant. The rod in (a) is irradiated for 2 s in a spot near the top (white circle indicates location and diameter of laser spot), resulting in a bend near the top of the rod in (b). After 2 min in darkness it relaxes to its former configuration (c). This cycle is then continued with irradiation in one spot (d,f), where the bend is indicated by the white circle, followed by relaxation (e,g). Scale bar ¼ 15 mm.
ß 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2009, 5, No. 24, 2902–2909
Bending Motions in Molecular Crystal Nanorods
Figure 4. Single 9AC nanorod moving in stepwise fashion as a result of repeated irradiation. a) The top portion of the rod, enclosed by the white ellipse, has been irradiated and pivots to the left (b), where the outlined ghost image shows the position of the rod from (a). The top portion of the rod is irradiated again (c) and as it relaxes, the top of the rod pivots towards the left again (d). e) The bottom portion of the rod is irradiated resulting in the bottom of the rod pivoting towards the left in (f). The stationary rod at left serves as a reference point for the motion. Scale bar ¼ 10 mm.
2.3.3. Quantification of Magnitude and Rate of Rod Bending Figures 3–5 show how 2PE can be used to induce bending dynamics in single 9AC nanorods. We now turn to an analysis of quantities like angles and rates of rod bending under illumination (kbend) and recovery in the dark (kunbend). All the rods exhibited exponential rises followed by a constant plateau. Figure 6a shows a plot of bend angle versus illumination time averaged over a population of rods from the same preparation batch. This data yields an average maximum bend angle of 248 and a bend rate, kbend ¼ 0.81 s1. There is a large distribution of rates in the bending data, however, and when individual curves are analyzed we find that these times can vary by a factor of 10 (0.3 s1 to 3.2 s1). In other words, the averaged data in Figure 6a provides only an approximate guide to the bending rate for an individual rod. Once the light was turned off, the rods straightened out again,
a 25
20
Bend Angle (deg)
as the 200-nm-diameter rods. Figure 5 shows a reversible bend induced in a single 35-nm rod under the same exposure conditions as the 200-nm-diameter rods.
15
10
5
0 0
1
2
3
4
5
6
7
8
Irradiation Time (s)
b 25
Bend Angle (deg)
20
15
10
5
0 0
100
200
300
400
500
Recovery Time (s)
Figure 5. Single 9AC rod 35 nm in diameter exposed to 800-nm light under the same conditions used for bending the 200-nm rods. In the left frame, the rod is at rest. The middle frame shows the rod after a 2-s illumination of the region enclosed by the black circle, resulting in a 178 bend. After 2 min in darkness,the rod revertsback to its originalshape in the right frame. Scale bar ¼ 5 mm. small 2009, 5, No. 24, 2902–2909
Figure 6. a) Average bending curve for 9AC rods excited at constant intensity in a single exposure lasting several seconds. This curve was generated by averaging data points acquired from bending runs on a sample of template-treated 9AC rods. Bending runs that terminated beforetheotherrunswereextrapolated asa flatlineatthefinalbend angle for each rod. b) The average recovery curve of the rods begins after irradiation ceases. This curve was generated by averaging data points acquired from runs of bending on a similar sample of rods. Fit lines shown are generated using Equations (3) and (4) for the bend and recovery curves, respectively.
ß 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
2905
full papers
C. J. Bardeen et al.
and the kunbend values showed much less variation. Figure 6b shows the decay curve averaged over different rods, along with the exponential fit with kunbend ¼ 0.0090 s1. In this case, the kunbend rates for individual rods only ranged from 0.0084 s1 to 0.0095 s1, or a factor of 1.13. Furthermore, the unbending followed the same decay law in the dark as the fluorescence recovery, indicating that it is molecular in origin. To explain the large variation in kbend rates, we consider three possible origins. The first is that different rods have different chemical characteristics (crystallinity, defect density, impurity levels) that determine the bending rate. Inspection of rod samples under TEM failed to reveal observable variations in crystallinity. Furthermore, morphology effects that lead to variations in kbend would be expected to affect kunbend as well. Since the distribution of kunbend values was much narrower than that of the kbend values, there is no independent evidence for variations in rod structure that would affect the deformation rates. A second possibility is that variations in laser intensity cause different excitation rates and thus different bending rates. The rods were confined to a 1-mm-thick well for observation, which is much shorter than the Rayleigh range of the focused laser beam. For each rod, the focus was optimized manually within the sample before staring the bending measurements. While the optimum focusing position could be determined, the quality of the focus may have varied due to imperfect focusing, local variations in scattering, or the proximity of dielectric interfaces. The nonlinear dependence of kbend on intensity described below in Equation (5a) would enhance the effects of small intensity variations. The last explanation for the observed variation in kbend is the presence of different amounts of adhesion to the glass surface by different rods. The evidence for surface effects includes the observation that uncoated rods exhibited bending behavior much less frequently than the SDScoated rods, and that individual rods were observed to bend more rapidly during the second and subsequent bending sequences. Taken together, these observations suggest that the variations in kbend originate from the last two factors, which are due to experimental nonidealities instead of variations in rod structure. The fastest observed kbend values probably provide a lower bound for the intrinsic kbend value of free rods.
3. Discussion 3.1. Molecular Origins of Reversible Bending Our earlier work on 9-tertbutyl-anthroate (9TBAE) showed that nanorods composed of this compound could undergo a nonreversible photodimerization that involved a crystal-to-crystal transition.[15] It is reasonable to assume that the 9AC photodimerization also proceeds in a crystal-to-crystal fashion. The bending is most likely due to the interface between reacted and unreacted 9AC crystal regions in the rods. This reasoning is supported by the observation that illumination of an entire segment results in asymmetric bending while illumination at a single point gives rise to a symmetric bend. Point illumination gives rise to two interfaces between reacted and unreacted molecules, resulting in two bends around the central point. Illumination of a segment from one end to a middle point, as shown in Figure 4, leaves only one interface
2906 www.small-journal.com
where the bend can occur. The bending then appears to be asymmetric since only one part of the rod moves while the other remains stationary. The physical origin of the bending could be strain arising from the lattice mismatch between unreacted and partially reacted regions in the rod. If the strain in any one region along the rod is proportional to the fraction of reacted molecules in that region,[29] then the bend angle should also be linearly proportional to the fraction of reacted molecules. A somewhat related mechanism posits that the crystal lattices of the monomer and dimer are sufficiently similar that the bend angle can vary smoothly with concentration of dimers within a specific region. In this case, the bend would be due to a structural effect, rather than to strain. In either case, we assume that the bend angle can vary continuously with the fraction of photoinduced dimers. If this assumption holds, then it is straightforward to generate a simple kinetic model for the lightinduced bending. For 2PE, the monomer and dimer populations Nmon and Ndim are governed by the following two differential equations: @Nmon ¼ 2dI 2 ’dim Nmon þ 2kdissoc Ndim @t
(1a)
@Ndim ¼ þdI 2 ’dim Nmon kdissoc Ndim @t
(1b)
where d is the two-photon absorption cross section, I is the light intensity, fdim is the quantum yield of photodimerization, and kdissoc is the rate of dimer dissociation back into monomers. The kinetics of reactive molecules in 1D chains (the 1D stacks of 9AC molecules) are more complicated than reflected in Equations (1a) and (1b) but can be approximated by assuming that a fraction P of the monomers remains after completion of the reaction. The existence of P is due to the fact that in a chain of reactive molecules, each molecule can react with either its left- or right-hand neighbor. As the reaction proceeds, some molecules find that both neighbors have already reacted, leaving the central molecule without a partner. In the ideal case of equal reaction probabilities between all molecules, it has been shown that P ¼ 0.13.[30,31] Note that conservation of the total number of molecules 0 means that Nmon þ 2Ndim ¼ Nmon and Equations (1a) and (1b) are not independent. This means that the kinetics can be completely described in terms of a single variable, which we define as X¼
2Ndim ¼ fraction of converted monomers 0 Nmon
(2)
As described above, the bend angle is proportional to the induced strain, which is proportional to X. a is defined to be the proportionality constant. During illumination, we neglect kdissoc and find ubend ¼ aX ¼ að1 PÞ 1 exp 2dI 2 ’dim t
(3)
If the illumination is then turned off, I ¼ 0 and we can no longer neglect the kdissoc term, resulting in a new time-
ß 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2009, 5, No. 24, 2902–2909
Bending Motions in Molecular Crystal Nanorods
dependence for the bend angle: ubend ¼ aX ¼ aX ðt ¼ 0Þ exp½kdissoc t
(4)
Equations (3) and (4) predict an exponential increase in bend angle under illumination followed by an exponential decay after the excitation light is turned off. We observe both phenomena experimentally. Equations (3) and (4) also predict values for the experimentally observed kbend and kunbend rates: kbend ¼ 2dI 2 ’dim
(5a)
kunbend ¼ kdissoc
(5b)
The value for kunbend is similar to the recovery time of the rod fluorescence after photoreaction, which should reflect kdissoc since dissociation results in a recovery of the strong monomer fluorescence. The value for kbend involves more variables but we can make an order of magnitude estimation. If we take a rough value of d ¼ 1051 cm4 s1 for crystalline anthracene[22] and estimate fdim ¼ 0.1, and combine them with our experimental value of I ¼ 4.2 1028 photons per cm2 s1 (assuming a 150-fs pulsewidth and a 1.5-mm-diameter spot), using Equation (5a) we find kbend 1 s1, which is close to the experimental value of 0.81 s1. Note that for the case of 2PE in a 200-nm-diameter nanorod, attenuation of the exciting beam as it propagates through the rod should be negligible, since the characteristic length for two-photon absorption is estimated to be greater than 1 cm due to the low d value. The calculated kbend value could change by at least an order of magnitude in either direction due to uncertainties in d, fdim, and I, but serves to demonstrate that for reasonable values of these parameters, we can calculate a value for kbend in good agreement with the experimental value. This indicates that Equations (5a) and (5b) can be used as a starting point to analyze the bending dynamics.
3.2. Limitations on Control of Nanorod Bending The results presented above demonstrate that it is possible to control both the location (by laser spot placement) and magnitude (by laser intensity and/or exposure time) of the induced bending. One aspect we were unable to control was the direction of the bend. Attempts to vary the bend direction by adjusting the position of the spot with respect to the rod axis were unsuccessful; a given rod always bent in the same direction, no matter what illumination conditions were used. The bend direction was consistent along a given rod as well, as can be seen from Figure 3. This observation suggests that the bend direction is determined by an intrinsic property of the nanorod, as opposed to an environmental factor like surface adhesion or irradiation profile. While defects may play a role, it is unlikely that defects would induce the same behavior at different points along a 60-mm-long rod. A more reasonable explanation involves the anisotropy of the 9AC crystal in the nanorod. Powder X-ray diffraction and TEM measurements have confirmed the preferential orientation of the 9AC crystal along the long axis of the nanorod.[25] However, knowing the orientation of the crystal with respect to the rod’s long axis still small 2009, 5, No. 24, 2902–2909
leaves an ambiguity as to the axial rotation of the crystal within the cylindrical rod. For example, if the c-axis of the crystal corresponds to the long axis of the rod, the angles of the a- and baxes within a specific rod lying on the surface are unknown. In general, these axes will be randomly oriented. In our simple example, if the preferred bend direction is along the a-axis, then the bend direction within a given rod will always be the same (along the a-axis) but will be different for different rods, since some will lie with the a-axis pointing to the left while others lie with the a-axis pointing to the right. The random orientation of the bend direction on the surface may also explain the observed variability in bend magnitude, since if the bend direction is not oriented parallel to the surface, the amount of observable bend angle in the plane of observation will decrease. Finally, it should be noted that although numerous methods have been reported for aligning nanowires with respect to their long axis, we were unable to find experiments where the rotational angle around its long axis can be controlled. If the intrinsic nanorod crystal orientation determines the bend direction, this would be an important difference with respect to liquid-crystal elastomers doped with photochromic molecules. In those types of structures, the bend direction is determined by a combination of extrinsic factors, like asymmetry in the light absorption, as well as by intrinsic properties like liquid crystal alignment.[29,32–34] For example, in azobenzene-doped elastomers, changing the polarization and intensity of light affords some control over the deformation but require that the light intensity varies within the structure. This intensity variation is usually due to Beer’s Law attenuation of the exciting beam, but this attenuation becomes negligible when nanometer-scale structures are used. For molecular crystal nanorods, on the other hand, there would not be a size limit for directional bending, as long as good crystallinity is maintained.
4. Conclusion We have expanded our earlier work on the reversible photoinduced bending of 9AC molecular crystal nanorods. We have shown that highly localized 2PE can induce bending at specific locations along a single rod, and that the phenomenon can be scaled from 200- to 35-nm-diameter rods. Furthermore, we have developed a simple kinetic model that can quantitatively describe the bending and relaxation behavior of individual rods. Remaining challenges include controlling the axial orientation of the rods in order to control the bending direction, and better control of rod–surface interactions to obtain more reproducible bend magnitudes and rates. Despite these challenges, this work shows that photomechanical effects in molecular crystal nanorods can be controlled in a reproducible and precise manner, and thus these materials have promise for applications in the area of nanomanipulation.
5. Experimental Section Preparation of 9AC nanorods: The 9AC nanorods were prepared using a template synthesis method that employs
ß 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
2907
full papers
C. J. Bardeen et al.
commercially available anodized alumina filters (Whatman Anodisc-13, 200-nm pore diameter, membrane thickness 60 mm, pore density 109 cm2).[16,35] To aid in the isolation of the nanorods, the templates were pretreated with a surfactant (SDS). A drop of SDS solution (1 wt%), acidified to pH 1.8 by addition of sulfuric acid, was deposited on both sides of the template and allowed to dry for 18 h. Then a solution of 13 mg 9AC (Aldrich, 99% purity), in 0.2 mL tetrahydrofuran (THF; HPLC grade 99.9% purity) was slowly deposited on both sides of the Anodisc. The template was then placed on a plastic stand in a glass jar containing filter paper soaked with additional THF (2 mL). The jar was sealed except for two holes poked in the lid of the jar that allowed air to flow at a slow rate through a gas bubbler (about 1 bubble per 3 s) to slowly evaporate the THF. After standing at room temperature for 36 h, the template was removed and polished with 1500 grit sandpaper to remove excess 9AC. The template was then dissolved in a solution of phosphoric acid (50% aqueous) creating a suspension of 9AC rods. To 2 mL of this suspension, an additional 0.4 mL SDS solution (1 wt% aqueous) was added to produce the final rod solution used for bending experiments. To make 35-nm rods, different alumina templates (Synkera OA-35-50-13 AAO, 35-nm pore diameter, membrane thickness 50 mm, pore density 1010 cm2) were employed and 9AC (6 mg in 0.1 mL THF) was deposited on both sides of the template. Observation of nanorod bending: To observe nanorod bending, 1-mm-deep well plates were made by coating microscope cover slips with a layer of gold 1-mm in thickness with a 1 cm2 square masked off in the middle. A drop of nanorod solution was deposited in the well and sandwiched by placing another cover slip on top. An inverted optical microscope equipped with phase contrast rings (Olympus IX70) and a 60 (1.2 NA) oil immersion objective was used for imaging the sample. To excite the rods, 150 fs 800-nm pulses generated by a Ti:sapphire laser (Spectra Physics-Mai Tai) at a 82-MHz repletion rate were focused onto the sample from below, resulting in a spot 1.5 mm in diameter (full width at half maximum). The peak intensity during each femtosecond pulse is 1010 W cm2, typical of intensities used for two-photon fluorescence imaging and well below the damage threshold of the crystal. For typical runs, bending was achieved by exposing a segment of the rod to the laser at a power of 100 mW for several seconds. To monitor the time resolved bending of the rods, rods were continuously monitored (at a rate of 30 frames per second) during and after laser exposure. Measurement of two-photon absorbtion of nanorods: To measure the 2PE spectrum of the 9AC nanorods, an aqueous rod suspension was placed on a microscope cover slip and imaged in the microscope with a 20 objective. A photomultiplier tube (Hamamatsu R5600U) was coupled to the camera port on the microscope to collect the fluorescence signal produced by exciting a bundle of nanorods at different wavelengths across the tunable range of the Mai-Tai laser (710–920 nm, 10-nm steps). A chopper (Stanford Research Systems SR 540) coupled with a Lock-in amplifier (SR 830 DSP) were used to detect the fluorescence signal, which was proportional to the intensity squared.
2908 www.small-journal.com
Acknowledgements Discussions with Prof. Rabih O. Al-Kaysi are gratefully acknowledged. We thank Dexter Humphrey and the Center for Nanoscale Science and Engineering (CNSE) at UCR for the use of the electron-beam evaporator. Electron microscopy measurements on the nanorods were performed at the Central Facility for Advanced Microscopy and Microanalysis (CFAMM) at UCR. This work was funded by the National Science Foundation, grant DMR-0907310.
[1] T. Ikeda, J. Mamiya, Y. Yu, Angew. Chem. Int. Ed. 2007, 46, 506– 528. [2] H. Finkelmann, E. Nishikawa, G. G. Pereira, M. Warner, Phys. Rev. Lett. 2001, 87, 015501–015504. [3] M. Yamada, M. Kondo, R. Myasato, Y. Naka, J. Mamiya, M. Kinoshita, A. Shishido, Y. Yu, C. J. Barrett, T. Ikeda, J. Mater. Chem. 2009, 19, 60–62. [4] Y. Yu, M. Nakano, T. Ikeda, Nature 2003, 425, 145. [5] M. Camacho-Lopez, H. Finkelmann, P. Palffy-Muhoray, M. Shelley, Nat. Mater. 2004, 3, 307–310. [6] M. Yamada, M. Kondo, J. Mamiya, Y. Yu, M. Kinoshita, C. J. Barrett, T. Ikeda, Angew. Chem. Int. Ed. 2008, 47, 4986–4988. [7] T. J. White, N. V. Tabiryan, S. V. Serak, U. A. Hrozhyk, V. P. Tondiglia, H. Koerner, R. A. Vaia, T. J. Bunning, Soft. Mater. 2008, 4, 1796– 1798. [8] K. D. Harris, R. Cuypers, P. Scheibe, C. L. v. Oosten, C. W. M. Bastiaansen, J. Lub, D. J. Broer, J. Mater. Chem. 2005, 15, 5043–5048. [9] M. Irie, S. Kobatake, M. Horichi, Science 2001, 291, 1769–1772. [10] I. Colombier, S. Spagnoli, A. Corval, P. L. Baldeck, M. Giraud, A. Leaustic, P. Yu, M. Irie, J. Chem. Phys. 2007, 126, 011101/1– 011101/3. [11] S. Kobatake, S. Takami, H. Muto, T. Ishikawa, M. Irie, Nature 2007, 446, 778–781. [12] K. Uchida, S. Sukata, Y. Matsuzawa, M. Akazawa, J. J. D. d. Jong, N. Katsonis, Y. Kojima, S. Nakamura, J. Areephong, A. Meetsma, B. L. Feringa, Chem. Commun. 2008, 326–328. [13] D. J. Flannigan, V. A. Lobastov, A. H. Zewail, Angew. Chem. Int. Ed. 2007, 46, 9206–9210. [14] D. J. Flannigan, P. C. Samartzis, A. Yurtsever, A. H. Zewail, Nano Lett. 2009, 9, 875–881. [15] R. O. Al-Kaysi, A. M. Muller, C. J. Bardeen, J. Am. Chem. Soc. 2006, 128, 15938–15939. [16] R. O. Al-Kaysi, C. J. Bardeen, Adv. Mater. 2007, 19, 1276–1280. [17] R. O. Al-Kaysi, R. J. Dillon, J. M. Kaiser, L. J. Mueller, G. Guirado, C. J. Bardeen, Macromolecules 2007, 40, 9040–9044. [18] D. K. Bucar, L. R. MacGillivray, J. Am. Chem. Soc. 2007, 129, 32–33. [19] S. Takahashi, H. Miura, H. Kasai, S. Okada, H. Oikawa, H. Nakanishi, J. Am. Chem. Soc. 2002, 124, 10944–10945. [20] Y. Ito, H. Fujita, J. Org. Chem. 1996, 61, 5677–5680. [21] J. Wolf, G. Hohlneicher, Chem. Phys. 1994, 181, 185–208. [22] I. Webman, J. Jortner, J. Chem. Phys. 1969, 50, 2706–2716. [23] W. J. Yang, M. S. Seo, X. Q. Wang, S. J. Jeon, B. R. Cho, J. Fluoresc. 2008, 18, 403–411. [24] A. Bree, R. Zwarich, Chem. Phys. 1993, 170, 185–192. [25] R. O. Al-Kaysi, R. J. Dillon, L. Zhu, C. J. Bardeen, J. Colloid Interface Sci. 2008, 327, 102–107. [26] T. Saitoh, K. Taguchi, M. Hiraide, Anal. Chim. Acta 2002, 454, 203– 208.
ß 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2009, 5, No. 24, 2902–2909
Bending Motions in Molecular Crystal Nanorods
[27] C. Pan, L. Zhang, J. Zhu, J. Luo, Z. Cheng, C. Wang, Nanotechnology 2007, 18, 015302/1–015302/6. [28] E. M. Purcell, Am. J. Phys. 1977, 45, 3–11. [29] M. L. Dunn, J. Appl. Phys. 2007, 102, 013506/1–013506/7. [30] E. R. Cohen, H. Reiss, J. Chem. Phys. 1963, 38, 680–691. [31] J. Even, M. Bertault, J. Chem. Phys. 1999, 110, 1087–1096. [32] M. Warner, L. Mahadevan, Phys. Rev. Lett. 2004, 92, 134302/1–134302/4.
small 2009, 5, No. 24, 2902–2909
[33] D. Corbett, C. L. v. Oosten, M. Warner, Phys. Rev. A 2008, 78, 013823/1–013823/4. [34] C. L. v. Oosten, D. Corbett, D. Davies, M. Warner, C. W. M. Bastiaansen, D. J. Broer, Macromolecules 2008, 41, 8592–8596. [35] R. O. Al-Kaysi, C. J. Bardeen, Chem. Commun. 2006, 1224–1226.
ß 2009 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Received: May 26, 2009 Published online: October 23, 2009
www.small-journal.com
2909
